Objective Quality Assessment Method for Stitched Images

Transcription

1 1 : (Meer Sadeq Billah et al.: Objective Quality Assessment Method for Stitched Images) (Special Paper) 232, (JBE Vol. 23, No. 2, March 2018) ISSN (Online) ISSN (Print) a), a) Objective Quality Assessment Method for Stitched Images Meer Sadeq Billah a) and Heejune Ahn a) (Field of View). 20,.,. Delaunay. -,, PSNR.. Abstract Recently, stitching techniques are used for obtaining wide FOV, e.g., panorama contents, from normal cameras. Despite many proposed algorithms, the no objective quality evaluation method is developed, so the comparison of algorithms are performed only in subjective way. The paper proposes a Delaunay-triangulation based objective assessment method for evaluating the geometric and photometric distortions of stitched or warped images. The reference and target images are segmented by Delaunay-triangulation based on matched points between two images, the average Euclidian distance is used for geometric distortion measure, and the average or histogram of PSNR for photometric measure. We shows preliminary results with several test images and stitching methods for demonstrate the benefits and application. Keyword: Quality assessment, objective measure, image stitching, panorama, reference based a) (Seoul National University of Science and Technology, Department of Electrical and Information Engineering) Corresponding Author : (Heejune Ahn) heejune@seoultech.ac.kr Tel: ORCID: ( )., (Ministry of Science and ICT, IITP). ( 360VR ( ) This work was supported by InstituteforInformation& communications TechnologyPromotion(IITP) grant funded by the Koreagovernment(MSIT) (No ,Moving Free-viewpoint 360VR Immersive Media System Design and Component Technologies). Manuscript received December 29, 2017; Revised Macrch 8, 2018; Accepted March 8, Copyright 2016 Korean Institute of Broadcast and Media Engineers. All rights reserved. This is an Open-Access article distributed under the terms of the Creative Commons BY-NC-ND ( which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited and not altered.

2 (JBE Vol. 23, No. 2, March 2018). 20., ,,,.. (FOV: Field of View) 60., (stitching) (mosaic) [1]., 360 MPEG [2].,,.,. homography [3,4]. 360, homography, [5].,.,, (foreground object) (ghost). 3, [1,3,6].,,, 3. homography, [7]. [8,9].,.,,.,.,.,..,.., Qureshi [10].,,... II

3 1 : (Meer Sadeq Billah et al.: Objective Quality Assessment Method for Stitched Images). III homography. IV..., Delaunay. 1. Delaunay N,, Fig 1,.., right center right (reference) stitched Reference (center) 1. () Fig 1. Proposed Image distortion measure of stitched images based on additional reference image (center).,.,..,.,.,.. 1) { }, { })., * * 2). ( Euclidean distance.) SIFT SURF, FAST, FAST (Features from Accelerated Segment Test) ORB (Oriented FAST and Rotated BRIEF). K-Nearest Neighbor. Delaunay [13]. Delaunay OpenCV Subdiv2D API.

4 (JBE Vol. 23, No. 2, March 2018) 2. (GD: Geometric distortion)., ( ( ). Fig 4 affine PSNR. PSNR.. 2. Fig 2. Workflow of proposed quality assessment PSNR( ) = PSNR(AffineTf(Stitched Triangle) - ReferenceTiangle 3. D- ( :, : ) Fig 3. Triangle matches examples between stitched (red) and reference (black) image Transform 4. PSNR Fig 4. Segment PSNR calculation based on affine transform matching of stitched to reference triangle

5 1 : (Meer Sadeq Billah et al.: Objective Quality Assessment Method for Stitched Images) III.. Fig 6 PSNR. PSNR., homography AutoStitch [4]. AutoStitch OpenCV Matlab.,,. [11,12] Fig 5. Fig Local segmented PSNR Fig 6. PSNR maps for test set 1 Left Center Right Test Set1 Test Set2 Test Set 3 5. (, ( ), ) Fig 5. Test image sets (left, reference center, and right)

6 (JBE Vol. 23, No. 2, March 2018),.. Fig. 7 PSNR.. Set 3,. Set 1 Set 2 Set 1.,. 1. AutoStitch Table 1. The performance values using the proposed assessment method Test set Avg G.D. Avg.PSNR Set Set Set PSNR SSIM.,,,. 7. PSNR Fig 7. The sub-triangle PSNR histograms for test sets. PSNR Table 1, 5 (> 0.8 ). r 5 Likert x PSNR y..,,. 20,

7 1 : (Meer Sadeq Billah et al.: Objective Quality Assessment Method for Stitched Images)..,.,.,.., AutoStitch.. (References) [1] D. Ghosh, N. Kaabouch, A survey on image mosaicing techniques, Journal of Visual Communication and Image Representation, Vol. 34, No.1, pp. 1-11, January, [2] P. Topiwala, W. Dai, M. Krishnan, A. Abbas, A. S. Doshi, D. Newman, Performance comparison of AV1, HEVC, and JVET video codecs on 360 (spherical) video, Applications of Digital Image Processing XL, Vol , p , September, [3] R. Szeliski, Image alignment and stitching: A tutorial. Foundations and Trends, Computer Graphics and Vision, Vol 2, No. 1, pp [4] Brown, M., & Lowe, D. G. (2007). Automatic panoramic image stitching using invariant features. International journal of computer vision, 74(1), [5] R. Hartley, A Zisserman, Multiple View Geometry in Computer Vision, 2nd Ed., Cambridge Press, New York, NY, USA. February, [6] F. Dornaika, R. Chung, Mosaicking images with parallax, Signal Processing: Image Communication, Vol. 19, No. 8, pp , [7] A. Eden, M. Uyttendaele, R. Szeliski, Seamless image stitching of scenes with large motions and exposure differences. In Computer Vision and Pattern Recognition, New York, NY, USA, pp [8] J. Zaragoza, T. J. Chin, M. S. Brown, D. Suter, As-projective-as-possible image stitching with moving DLT, Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Portland, Oregon, USA, pp , 2013 [9] C. H. Chang, Y. Sato, Y. Y. Chuang. Shape-preserving half-projective warps for image stitching, Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Columbus, Ohio, USA pp , [10] Qureshi, H. S., Khan, M. M., Hafiz, R., Cho, Y., & Cha, J. (2012). Quantitative quality assessment of stitched panoramic images. IET Image Processing, 6(9), [11] Wu, Y. test Image for Image Stitching, available on com/ppwwyyxx/openpano [12] V. Pham, test images for Image stitching, available on com/phvu/misc/tree/master/imagestitch [13] Delaunay, Boris. "Sur la sphere vide." Izv. Akad. Nauk SSSR, Otdelenie Matematicheskii i Estestvennyka Nauk (1934): 1-2. (Meer Sadeq Billah) : BGC Trust University Bangladesh (Computer Science & Information Technology) (Bachelor of Science) ~ 2008 : Eastern Bank Limited, Bangladesh (Analyst, Service Management) ~ 2012 : SurroundApps Inc., Bangladesh (Sr. Software Engineer) ~ 2015 : Samsung R&D Institute Bangladesh (Technical Lead) : Seoul National Univ. of Science & Technology (Electrical & Information Engineering) (Master of Science) - ORCID : - : computer vision, video processing, computer graphics

8 (JBE Vol. 23, No. 2, March 2018) KAIST ( ) ( ) ~ 2002 : LG ( ) ~ 2003 : TmaxSoft ( ) ~ : ( ) - ORCID : - :,,